view full paper

advertisement
ICAS 2000 CONGRESS
ERROR BUDGETING AND THE DESIGN OF LARGE
AEROSTRUCTURES
R. Odi*; G. Burley; S. Naing; A; Williamson; J. Corbett
B42A, School of Industrial and Manufacturing Science, Cranfield University, Cranfield,
MK43 0AL, England
*e-mail: a.r.a.odi@cranfield.ac.uk
Keywords: error budgeting, design methods, jigless assembly, aerostructures
Abstract
Traditionally the manufacture of both military
and civil aircraft has relied on the use of fixed
tooling designed specifically for individual
product types (ECATA 1995 [1]). Coproduction of aircraft is resulting in demands
for higher standards of manufacturing quality to
ensure that parts and sub-assemblies from
different countries are compatible and
interchangeable. As a results, the existing
manufacturing and assembly methods using
large numbers of dedicated jigs and tools is now
seen as being commercially undesirable, and
technically flawed (Burley and Corbett, 1998
[2]).
Error budgeting has been succesfully used
for the design of precision machines [3, 4].
According to Donaldson [5], “… an error
budget is a system analysis tool, used for
prediction and control of the total error of a
system at the design stage for systems where
accuracy is an important measure of
performance”.
The Jigless Aerospace Manufacture (JAM)
project is a three-year project supported by the
UK Engineering and Physical Sciences
Research Council (EPSRC) and aims to
investigate
the
significant
scientific,
technological and economic issues to enable a
new design, manufacture and assembly
philosophy based on eliminating or minimising
product specific jigs, fixtures and tooling.
As part of this on-going research
programme, it proposed to investigate the use of
error budgeting as a tool to enable the design of
large aerostructures which would be easily
assembled without recourse to expensive jigs
and fixtures. This technique has the potential to
allow the comparison of several concepts and
configurations early in the design process and
so help the selection process as well as
highlighting areas where redesign should be
considered at the detail design stage.
This paper will present the results of the
investigation, assessing its potential and limits
as well as its suitability for the design of large
aerospace structures.
1 Introduction
Error budgeting has been successfully used for
the design of precision machines. According to
Donaldson [5], “… an error budget is a system
analysis tool, used for prediction and control of
the total error of a system at the design stage for
systems where accuracy is an important
measure of performance”.
Most of the literature on error budgeting is
related to the design of machines, in particular
machine tools [6] and co-ordinate measuring
machines (CMMs). The other main source of
literature on the subject is related to the design
of optical systems (such as telescopes etc.).
Error Budgeting is used in machine design
as an analytical tool to predict the total error of
the machine system at the design stage. The
starting point is a target error for the total
machine which is then distributed through an
error budget to set the individual sub-systems
errors [7].
592.1
R. Odi, G. Burley, S. Naing, A. Williamson, J. Corbett
The error budget is then used as a trade-off
tool between the various sub-systems of the
machine to achieve a balance between the target
and the difficulties in achieving it. Error
budgeting thus provides a systematic method to
determine the degree of difficulty in achieving a
particular total error target.
Error budgets are used at all stages of the
machine development process: conceptual
design, detail design, planning, prototype
development, build, series production etc.
The main idea for the use of error budgets
in the design of aero-structures is to categorise
as errors any imperfections or variations on
important features which are used to put
together components into sub-assemblies and
assemblies. This analytical technique relies on
the concepts of key characteristics (KC) [8] and
datum flow chains (DFC)[9] and featurised
DFC.
2 Methodology
The Error Budgeting technique rests on three
main pillars: KC methodology, DFC and
featurised DFC diagrams and the application of
geometric dimensioning and tolerancing
symbols and terminology. Three building
blocks, which are at the centre of the technique,
are presented briefly.
Sub-Assembly Level
PKC #1
Component Level
PKC #1.1
A
B
PKC #1.2
SUB-ASSEMBLY AND COMPONENTS
PKC #1
Total Thickness
PKC #1.1
Part A
Thickness
PKC #1.2
Part B
Thickness
KC FLOWDOWN
Figure 1 KC Flowdown Example
2.1 Key Characteristics
As part of the design process, the integrated
product team (IPT) or design build team (DBT)
produces a series of important product features
whose variation significantly affect the product
function, form and quality10. These features are
called key characteristics (KCs). Product KCs
(PKCs) are the main KC category and they flow
down through the system to component level.
For the JAM project, once the product KCs have
been identified, the means by which they are
realised are divided into two KC categories:
assembly KCs (AKCs) and manufacturing KCs
(MKCs) for PKC realisation by assembly and
manufacturing processes respectively.
In the example shown in Figure 1, a twopart sub-assembly, the main PKC is the overall
thickness of the joint. This flows down to two
PKCs at the component level: the thickness of
the two plates. For a given manufacturing
strategy, a number of processes will be required
to create the plate thickness. Those
manufacturing processes, which significantly
affect the realisation of the plate thickness, will
be identified as MKCs. For a given assembly
strategy, certain processes will significantly
impact the realisation of the overall joint
thickness: these processes become AKCs.
Once the KCs have been identified, a clear
strategy must be defined to determine how they
will be realised. It is important that at the
earliest opportunity attention is given to the way
in which the product will be assembled. Datum
flow chains (DFC) have been created to
represent the underlying logic of an assembly at
592.2
ERROR BUDGETING AND THE DESIGN OF LARGE AEROSTRUCTURES
an abstract level. As such they can be used to
capture a product assembly strategy.
2.2 Datum Flow Chains
The DFC diagrams are part of a new approach
to conceptualise the design of complex
assemblies. It is now being accepted that not all
links between parts have the same importance
when it comes to assembly. By differentiating
those links which establish and transfer
dimensional location between parts (mates) and
those that are simply there for strength or
support (contacts), it is possible to represent
graphically the dimensional transfers for given
assemblies.
A datum flow chain is a directed acyclic
graph representation of the assembly with nodes
representing the parts and arcs representing
dimensional relationships between them.9
The initial work of Mantripragada and coworkers has been extended by Schwemmin11 to
incorporate the actual features that participate in
the assembly of the product. This gave rise to
the featurised DFC method which is intended to
assess the assemblability of a given assembly
concept via the inclusion of manufacturing
process capability data. Schwemmin’s method
is primarily geared towards existing products so
that assembly problems can be traced and better
understood.
Hence
the
reliance
on
manufacturing
process
capability
data.
Furthermore, variations that are due to the
rotation of parts are not accounted for by this
method. Nevertheless, the featurised DFC
method remains a useful and powerful tool for
analysing assembly, using a language accessible
to most engineers.
Error budgeting as proposed in this
document can be thought of as an adaptation of
Schwemmin’s featurised DFC method.
L2
L1
Feature 1 (F1)
Feature 2 (F2)
Part 2
Part 1
Part 1 DRF
Features:
A1, B1 and
C1
L3
DRF = Datum Reference Frame
Part 2 DRF
Features:
A2, B2 and
C2
Figure 2 Typical featurised DFC diagram
2.3 Use of GD&T Symbols and Terminology
The main similarity between error budgeting for
assembly and Schwemmin’s methodology lies
in the use of GD&T symbols and terminology
and the use of the DFC diagrams, which are
graphical representation of the underlying
assembly concepts for a given product. The
main difference is the way in which those
symbols are used. Also, in error budgeting for
assembly, design tolerances play a central part
to enable the comparison at the earliest stage of
given assembly concepts for a product.
Figure 3 Geometric characteristics categories [12]
There are five main categories of geometric
characteristics that can be used in error
budgeting to represent part feature variations:
form, profile, orientation, runout and location
controls. Form and profile controls are generally
tied to individual features and do not make
reference to datums. Orientation, runout and
location controls require in general reference to
datums.
592.3
R. Odi, G. Burley, S. Naing, A. Williamson, J. Corbett
The method is explained in more detail in
the following sections.
3 Error Budgeting for Assembly-Centric
Design
To use error budgeting to design assemblies,
there is a need to define a few concepts. There is
also a requirement to express the necessary
adaptation to the method as used by machine
tool and optical system designers.
3.1 Error in an Assembly
For an assembly of parts, an error is defined as
any variation from the idealised system.
Essentially these variations from perfection will
be captured by the application of tolerances to
ensure the proper function of the assembled
product. These variations will include tolerances
(dimensional & geometric) but also changes in
shape or form due to the environment
(temperature, moisture etc.).
3.2 Transfer Error and Mate Error
During the assembly of a product, a series of
parts or fixtures are used in such a way that
parts are positioned, secured and then fastened.
The use of DFC and featurised DFC diagrams
enable the designer to represent graphically the
way parts, fixtures or jigs are used to realise the
assembly. Due to the imperfect nature of these
elements, the dimensional transfer will be
subject to variation. If one considers only one
source of variation, namely geometric, the
errors that result from the assembly process can
be classified into transfer and mate errors.
Transfer errors are used to capture
variations due to the dimensional transfers
within the part and the imperfection of the
datum features themselves. Thus transfer errors
are a combination of datum form and profile
tolerances and assembly feature orientation and
location tolerances. If an assembly feature is
also part of the part datum reference frame, then
the feature orientation and location tolerances
are no longer relevant. Transfer errors are
applied to both the locating and located features.
Mate errors are a combination of the
dimensional, form, profile and runout tolerances
of the assembly features that establishes and
transfers dimensional location between the
parts. These mate variations are due to the
imperfections of the features themselves. Mate
errors are related both to the locating and
located features.
The definitions of transfer and mate errors
apply equally to fixtures and jigs. The
definitions are summarised in the following
table.
Datum
Feature
Ordinary
Feature
Transfer
Error
- form
tolerance, εf
- profile
tolerance, εp
- orientation
tolerance, e0
- location
tolerance, el
Mate Error
- dimensional
tolerance, ed
- form tolerance,
ef
- profile
tolerance, ep
- runout
tolerance, er
For a given feature-to-feature link, the
transfer error, Et, can be expressed as:
Et = f (ε f , ε p , eo , el )
(1)
The mate error, Em, is given by:
Em = g (ed , e f , e p , er )
(2)
The exact form of Equation 1 and Equation
2 will be ascertained in future work. They are
likely to be root sum square (RSS) statistical
additions because not all errors will be at their
maximum.
For a typical part-to-part assembly, such as
that in Figure 2, the error concepts introduced
above can be presented in a table as follows:
592.4
ERROR BUDGETING AND THE DESIGN OF LARGE AEROSTRUCTURES
Link
Error
type
Dat
um
L1
TRF
A
B
C
L2
MAT
Fea
ture
F1
F1
F2
Form
Profile
Orientation
FLT
x
x
x
STR
x
x
x
CIR
x
x
x
CYL
x
x
x
LPF
x
x
x
SPF
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Runout
PER
PAR
ANG
x
x
x
TRF = Transfer Error
Location
CRO
TRO
x
x
x
x
Dim
ensi
onal
±
POS
SYM
CON
x
x
x
DIM
x
x
MAT = Mate Error
Form
Profile
Orientation
Runout
Location
FLT = Flatness
STR = Straightness
CIR = Circularity
CYL = Cylindricity
LPF = Line Profile
SPF = Surface Profile
PER = Perpendicularity
PAR = Parallellism
ANG = Angularity
CRO = Circular Runout
TRO = Total Runout
POS = Position
SYM = Symmetry
CON = Concentricity
Table 2 Odinary feature error equations
Assuming RSS summation, the transfer error
(Equation 1) can be written as:
Et = ε 2f + ε p2 + eo2 + el2
(3)
Similarly, the mate error (Equation 2) is given
by:
Em = ed2 + e 2f + e 2p + er2
Error due to Equations
form
(eFLT )2 + (eSTR )2
ef =
2
2
+ (eCIR ) + (eCYL )
2
2
profile
e p = (eLPF ) + (eSPF )
runout
er =
(eCRO )2 + (eTRO )2
(10)
el =
(e POS )2 + (eSYM )2
2
+ (eCON )
(11)
N
∑ (e )
i =1
(12)
2
DIM i
N is number of toleranced dimensions
Equations
εf =
(ε FLT )2 + (ε STR )2
2
2
+ (ε CIR ) + (ε CYL )
For N number of features, the combined
error, eN, can be expressed as:
(5)
eN =
Equation 5
Profile
(9)
ef =
Table 1 Datum feature error equations
Error
due to
Form
eo =
location
dimension
(8)
(e PER )2 + (ePAR )2
2
+ (e ANG )
orientation
(4)
It is important to note that not every tolerance
type will be present for a given feature. Also the
error due to a particular geometric characteristic
will be a summation of different types of
tolerances within that group. For example,
errors due to form variation may include any
combination of flatness, straightness, circularity
and cylindricity tolerances.
(7)
εp =
(ε LPF )2 + (ε SPF )2
(6)
∑ (e
N
i =1
)
2
feature i
(13)
The initial total error for an assembly is
given by:
ΕA =
∑ (E )
2
t Li
+ ∑ (E m )Lj
2
(14)
i, j are DFC links
592.5
R. Odi, G. Burley, S. Naing, A. Williamson, J. Corbett
These equations can be used to create an
initial error budget that can be used to assess
various assembly concepts. The aim of the
design process is to find an assembly concept
with an acceptable error and cost level. The
larger the permitted error on each of the
features, the cheaper the cost of manufacture.
However this may result in fitting problems or
in the impossibility of delivering a given KC.
On the other hand, a reduction in error has some
cost implications. So the whole design team can
start making some trade-offs based on the error
budget. It is worth noting that if a part is overconstrained, it results in an increase of the total
product error. This is consistent with the fact
that over-constraint occurs when a number of
features compete for the elimination of the same
degrees of freedom. As such, they represent
additional sources of variation.
Once processes are better defined and the
design progresses, the error budget can be
refined further by resolving the errors along the
main datum reference frame (DRF) axes of the
assembly. An important part of the design
process is the assignment of datums for various
assemblies, sub-assemblies, components and
tooling. This aspect of the design process is
critical for the refinement of the error budget
because it enables the rough orientation of the
main components with respect to the product
DRF. Each error source can then be assessed
regarding its influence along the product DRF
axes.
4 Example: Bulkhead Assembly
The following example presents an error budget
for a bulkhead assembly. Two bulkheads need
to be put together using an assembly tool. The
main requirement to satisfy is the alignment of
the upper contours of the bulkhead with respect
to the assembly reference frame.
Figure 4 Bulkhead assembly [11]
Figure 4 shows the bulkhead assembly where
the two bulkheads BH1 and BH2 locate to the
tool details.
Figure 5 GD&T callouts of the assembly features [11]
592.6
ERROR BUDGETING AND THE DESIGN OF LARGE AEROSTRUCTURES
The callouts in Figure 5 refers to the design
intent (i.e. the values shown are tolerances not
manufacturing process capability data).
L13
L7
Loc1 Pln A
BH1 Sur IML
BH1
L8
L1
The Featurised DFC diagram is shown in Figure
6. The KC is delivered from secondary features
in each of the bulkheads and it is shown in the
picture. All the features and their callouts are
listed in the following table.
Loc1 Hol B
L2
L9
Loc1 Slot C
L3
Tool
L4
KC
Loc2 Pln A
L10
L5
Loc2 Hol B
L11
Loc2 Slot C
BH2 Sur IML
BH2
L6
L14
L12
Figure 6 Featurised DFC diagram for the proposed
assembly concept [11]
Table 3 Feature callouts
Part Feature
Tool Loc1 Pln A
Loc1 Hol B
Loc1 Slot C
Loc2 Pln A
Loc2 Hol B
Loc2 Slot C
Datum Pln A
Datum Pln B
Datum Pln C
BH1 BH1 Pln A
BH1 Hol B
BH1 Hol C
BH1 Sur IML
BH2 BH2 Pln A
BH2 Hol B
BH2 Hol C
BH2 Sur IML
Callout
SPF
POS
POS
SPF
POS
POS
FLT
FLT
FLT
FLT
PER
POS
SPF
FLT
PER
POS
SPF
Callout Type
Profile
Location
Location
Profile
Location
Location
Form
Form
Form
Form
Orientation
Location
Profile
Form
Orientation
Location
Profile
Tolerance
0.010
0.005
0.005
0.010
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.020
0.005
0.005
0.005
0.020
592.7
R. Odi, G. Burley, S. Naing, A. Williamson, J. Corbett
Table 4 Link-Feature Table
Link Type
Datum
Feature
Ordinary
Feature
L1
TRF
Datum Pln A
Datum Pln B
Datum Pln C
L2
TRF
Datum Pln A
Datum Pln B
Datum Pln C
L3
TRF
Datum Pln A
Datum Pln B
Datum Pln C
L4
TRF
Datum Pln A
Datum Pln B
Datum Pln C
L5
TRF
Datum Pln A
Datum Pln B
Datum Pln C
L6
TRF
Datum Pln A
Datum Pln B
Datum Pln C
L7
L8
L9
L10
L11
L12
MAT
MAT
MAT
MAT
MAT
MAT
L13
TRFKC*
BH1 Pln A
BH1 Hol B
BH1 Hol C
L14
TRFKC*
BH2 Pln A
BH2 Hol B
BH2 Hol C
Form
Tolerance
FLT
Profile
Tolerance
SPF
Location
Tolerance
POS
Orientation
Tolerance
PER
0.005
0.005
0.005
Loc1 Pln A
0.005
0.005
0.005
Loc1 Hol B
0.005
0.005
0.005
0.005
Loc1 Slot C
0.005
0.005
0.005
0.005
Loc2 Pln A
0.005
0.005
0.005
Loc2 Hol B
0.005
0.005
0.005
0.005
Loc2 Slot C
Loc1 Pln A
Loc1 Hol B
Loc1 Slot C
Loc1 Pln A
Loc1 Hol B
Loc1 Slot C
0.005
0.010
0.010
0.005
BH1 Sur IML
0.020
0.005
BH2 Sur IML
0.020
*
For transfer error links to feature that deliver directly a KC (such as L13 and L14 above) it is necessary to add any error due to the
feature itself. This adjustment is termed a KC error adjustment.
Table 5 Link Transfer Error Table
Link
Datum Features
Form
Profile
Ordinary Features
Location
Orientatio
n
KC Error
Adjustme
nt
Total
Transfer
Error
(Et)Li
L1
L2
L3
0.00866
0.00866
0.00866
0.0
0.005
0.005
0.0
0.0
0.0
0.00866
0.010
0.010
0.0
0.0
0.0
0.0
0.0
0.0
592.8
ERROR BUDGETING AND THE DESIGN OF LARGE AEROSTRUCTURES
L4
L5
L6
L13
L14
Datum Features
Ordinary Features
0.00866
0.00866
0.00866
0.005
0.005
0.0
0.005
0.005
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.020
0.020
0.00866
0.010
0.010
0.02061
0.02061
Table 6 Link Mate Error Table
Link
Ordinary Features
Form
Profile
Total
Mate
Error
(Em)Li
L7
L8
L9
L10
L11
L12
0.0
0.0
0.0
0.0
0.0
0.0
0.010
0.0
0.0
0.010
0.0
0.0
0.010
0.0
0.0
0.010
0.0
0.0
Using equation 14 (which is an RSS summation
of transfer and mate errors) and the data in
Table 5 and Table 6, the total error budget of the
assembly is 0.0274. The error budget for the KC
delivery is 0.040. The difference between the
two values comes from the L13 and L14 links,
which are omitted from the error budget for the
assembly.
5 Conclusion
This paper detailed the work undertaken on the
application of error budgeting to the design of
aerostructures. Several concepts have been
introduced that allow component variations to
be accounted for in the assessment of assembly
strategies.
Work is being carried out to ascertain the best
way to resolve the error component along
assembly main DRF axes. Such refinements to
the error budget will enable a better assessment
of assemblies.
The technique is has been applied to the JAM
demonstrator current assembly process to
provide a basis for comparison with the
proposed alternatives to the current build
process.
Acknowledgement
The authors would like to acknowledge the
financial support of the UK Engineering and
Physical Sciences Research Council and the
main industrial partners BAE SYSTEMS
(Military Aircraft & Aerostructures, Airbus UK
and Sowerby Research Centre), Bombardier
Shorts Aerospace and Comau UK Ltd
References
[1] ECATA. A Tooling Study to Reduce the Cost of
Aircraft Manufacture, Multinational Team Project,
Final Report, 1995.
[2] Burley, G; Corbett, J. Flyaway Tooling for Higher
Quality, More Cost-Effective Aerostructures, SAE
Paper No. 98AMTC-59, SAE Aerospace Manufacture
and Technology Conference, Seattle, June 1998.
[3] Thompson, D C; McKeown, P A; “The Design of an
Ultra-Precision CNC measuring Machine”, Annals of
the CIRP, Vol. 38/1, 1989, pp. 501-504.
[4] Thompson, D C; Fix, B L; “Comparison Between
Predicted and Actual Accuracies for an UltraPrecision CNC Measuring Machine”; Proceedings
8th International Precision Engineering Seminar,
Compiègne, France, 15-19 May 1998, 6 pp.
[5] Donaldson, R R; Patterson, S R; “Design and
Construction of a Large, Vertical Axis Diamond
Turning Machine”, Proc. SPIE, Vol. 433, Aug. 1983,
pp. 62-67.
592.9
R. Odi, G. Burley, S. Naing, A. Williamson, J. Corbett
[6] Homan, B S; Thornton, A C; “Precision Machine
Design Assistant: A Constraint-Based Tool for the
Design and Evaluation of Precision Machine Tool
Concepts”, Proceedings AIEDAM Spring ’98
Conference, 1998.
[7] Wills-Moren, W J; “Error budgeting in machine
design”, Lecture Notes, Cranfield University, Oct.
1993 –unpublished
[8] Thornton, A C; “A mathematical framework for the
Key Characteristics Concept”, Research in
Engineering Design, 1999, pp.
[9] Mantripangada, et al. “Assembly Orientated Design:
A new approach to designing assemblies”,
Proceedings 5th IFIP WG5.2 Workshop on Geometric
Modelling in Computer-Aided Design, 18-23 May
1996, Airlie, VA USA.
[10] Lee, D J; Thornton, A C; “Enhanced Key Characteristics
Identification Methodology for Agile Design”, Agile
Manufacturing Forum, March, 1996 Boston, MA.
[11] Schwemmin, R J; “Predicting Assembly Performance
with Featurized Datum Flow Chains”, Paper No.
1999-01-2271, SAE AMT Conference, 8-10 June
1999, Bellevue WA, USA.
[12] Meadows, J D; “Geometric Dimensioning and
Tolerancing”, Marcel Dekker Inc., New York, 1995.,
p. 56-57.
592.10
Download